(******************************************************************************)
INCLUDE "utils.ml"
+
+external init_lib : unit -> unit = "caml_init_lib"
+
+exception CPlusPlusError of string
+
+let () = Callback.register_exception "CPlusPlusError" (CPlusPlusError "")
+
+let () = init_lib ()
+
+
type tree
-type 'a node = int
+type 'a node = private int
type node_kind = [`Text | `Tree ]
-
-let compare_node : 'a node -> 'a node -> int = (-)
+
+external inode : 'a node -> int = "%identity"
+external nodei : int -> 'a node = "%identity"
+let compare_node x y = (inode x) - (inode y)
let equal_node : 'a node -> 'a node -> bool = (==)
-
-(* abstract type, values are pointers to a XMLTree C++ object *)
-external int_of_node : 'a node -> int = "%identity"
external parse_xml_uri : string -> int -> bool -> bool -> tree = "caml_call_shredder_uri"
external parse_xml_string : string -> int -> bool -> bool -> tree = "caml_call_shredder_string"
-external save_tree : tree -> string -> unit = "caml_xml_tree_save"
-external load_tree : string -> int -> tree = "caml_xml_tree_load"
+external tree_save : tree -> Unix.file_descr -> unit = "caml_xml_tree_save"
+external tree_load : Unix.file_descr -> tree = "caml_xml_tree_load"
external nullt : unit -> 'a node = "caml_xml_tree_nullt"
-let nil : 'a node = -1
-let root : [`Tree ] node = 0
+let nil : [`Tree ] node = nodei ~-1
+let nulldoc : [`Text ] node = nodei ~-1
+let root : [`Tree ] node = nodei 0
external text_get_tc_text : tree -> [`Text] node -> string = "caml_text_collection_get_text"
external text_is_empty : tree -> [`Text ] node -> bool = "caml_text_collection_empty_text"
-let text_is_empty t n =
- (equal_node nil n) || text_is_empty t n
-
-
+let text_is_empty t n = (equal_node nulldoc n) || text_is_empty t n
external text_is_contains : tree -> string -> bool = "caml_text_collection_is_contains"
external text_count_contains : tree -> string -> int = "caml_text_collection_count_contains"
external text_contains : tree -> string -> [`Text ] node array = "caml_text_collection_contains"
external text_unsorted_contains : tree -> string -> unit = "caml_text_collection_unsorted_contains"
external text_get_cached_text : tree -> [`Text] node -> string = "caml_text_collection_get_cached_text"
+
+external tree_root : tree -> [`Tree] node = "caml_xml_tree_root"
+external tree_subtree_size : tree -> [`Tree] node -> int = "caml_xml_tree_subtree_size"
+
+let tree_is_nil x = equal_node x nil
+external tree_parent : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_parent" "noalloc"
+external tree_parent_doc : tree -> [`Text ] node -> [`Tree ] node = "caml_xml_tree_parent_doc" "noalloc"
+(*external tree_prev_doc : tree -> [`Text ] node -> [`Tree ] node = "caml_xml_tree_prev_doc" "noalloc" *)
+external tree_first_child : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_first_child" "noalloc"
+external tree_first_element : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_first_element" "noalloc"
+external tree_tagged_child : tree -> [`Tree] node -> Tag.t -> [`Tree] node = "caml_xml_tree_tagged_child" "noalloc"
+external tree_next_sibling : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_next_sibling" "noalloc"
+external tree_next_element : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_next_element" "noalloc"
+external tree_tagged_sibling : tree -> [`Tree] node -> Tag.t -> [`Tree] node = "caml_xml_tree_tagged_sibling" "noalloc"
-external tree_serialize : tree -> string -> unit = "caml_xml_tree_serialize"
-
-external tree_unserialize : string -> tree = "caml_xml_tree_unserialize"
-
-external tree_root : tree -> [`Tree] node = "caml_xml_tree_root"
+external tree_prev_sibling : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_prev_sibling" "noalloc"
+external tree_is_leaf : tree -> [`Tree] node -> bool = "caml_xml_tree_is_leaf" "noalloc"
+external tree_last_child : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_last_child" "noalloc"
+external tree_is_first_child : tree -> [`Tree] node -> bool = "caml_xml_tree_is_first_child" "noalloc"
-let tree_is_nil x = equal_node x nil
-external tree_parent : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_parent"
-external tree_parent_doc : tree -> [`Text ] node -> [`Tree ] node = "caml_xml_tree_parent_doc"
-external tree_prev_doc : tree -> [`Text ] node -> [`Tree ] node = "caml_xml_tree_prev_doc"
-external tree_first_child : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_first_child"
-external tree_tagged_child : tree -> [`Tree] node -> Tag.t -> [`Tree] node = "caml_xml_tree_tagged_child"
-external tree_next_sibling : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_next_sibling"
-external tree_tagged_sibling : tree -> [`Tree] node -> Tag.t -> [`Tree] node = "caml_xml_tree_tagged_sibling"
-
-external tree_prev_sibling : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_prev_sibling"
-external tree_is_leaf : tree -> [`Tree] node -> bool = "caml_xml_tree_is_leaf"
-external tree_last_child : tree -> [`Tree] node -> [`Tree] node = "caml_xml_tree_last_child"
-external tree_is_first_child : tree -> [`Tree] node -> bool = "caml_xml_tree_is_first_child"
-
-(* external tag : tree -> [`Tree ] node -> T = "caml_xml_tree_tag"*)
-external tree_tag_id : tree -> [`Tree ] node -> Tag.t = "caml_xml_tree_tag_id"
+external tree_tag_id : tree -> [`Tree ] node -> Tag.t = "caml_xml_tree_tag_id" "noalloc"
let tree_is_last t n = equal_node nil (tree_next_sibling t n)
-external tree_prev_text : tree -> [`Tree] node -> [`Text ] node = "caml_xml_tree_prev_text"
+(*external tree_prev_text : tree -> [`Tree] node -> [`Text ] node = "caml_xml_tree_prev_text" "noalloc" *)
-external tree_my_text : tree -> [`Tree] node -> [`Text ] node = "caml_xml_tree_my_text"
-external tree_next_text : tree -> [`Tree] node -> [`Text ] node = "caml_xml_tree_next_text"
+external tree_my_text : tree -> [`Tree] node -> [`Text ] node = "caml_xml_tree_my_text" "noalloc"
+(*external tree_next_text : tree -> [`Tree] node -> [`Text ] node = "caml_xml_tree_next_text" "noalloc" *)
external tree_doc_ids : tree -> [`Tree ] node -> [`Text ] node * [`Text ] node = "caml_xml_tree_doc_ids"
-let text_size tree = int_of_node (snd ( tree_doc_ids tree (Obj.magic 0) ))
+let text_size tree = inode (snd ( tree_doc_ids tree root ))
-let text_get_cached_text t x =
- if x == -1 then ""
+let text_get_cached_text t (x:[`Text] node) =
+ if x == nulldoc then ""
else
text_get_cached_text t x
-external tree_text_xml_id : tree -> [`Text ] node -> int = "caml_xml_tree_text_xml_id"
-external tree_node_xml_id : tree -> [`Tree ] node -> int = "caml_xml_tree_node_xml_id"
-external tree_is_ancestor : tree -> [`Tree ] node -> [`Tree ] node -> bool = "caml_xml_tree_is_ancestor"
-external tree_tagged_desc : tree -> [`Tree ] node -> Tag.t -> [`Tree ] node = "caml_xml_tree_tagged_desc"
-external tree_tagged_foll_below : tree -> [`Tree ] node -> Tag.t -> [`Tree ] node -> [`Tree ] node = "caml_xml_tree_tagged_foll_below"
-external tree_subtree_tags : tree -> [`Tree ] node -> Tag.t -> int = "caml_xml_tree_subtree_tags"
+external tree_text_xml_id : tree -> [`Text ] node -> int = "caml_xml_tree_text_xml_id" "noalloc"
+external tree_node_xml_id : tree -> [`Tree ] node -> int = "caml_xml_tree_node_xml_id" "noalloc"
+external tree_is_ancestor : tree -> [`Tree ] node -> [`Tree ] node -> bool = "caml_xml_tree_is_ancestor" "noalloc"
+external tree_tagged_desc : tree -> [`Tree ] node -> Tag.t -> [`Tree ] node = "caml_xml_tree_tagged_desc" "noalloc"
+external tree_tagged_foll_below : tree -> [`Tree ] node -> Tag.t -> [`Tree ] node -> [`Tree ] node = "caml_xml_tree_tagged_foll_below" "noalloc"
+external tree_subtree_tags : tree -> [`Tree ] node -> Tag.t -> int = "caml_xml_tree_subtree_tags" "noalloc"
+type unordered_set
+external unordered_set_alloc : int -> unordered_set = "caml_unordered_set_alloc"
+external unordered_set_length : unordered_set -> int = "caml_unordered_set_length"
+external unordered_set_insert : unordered_set -> int -> unit = "caml_unordered_set_set" "noalloc"
-type int_vector
-external int_vector_alloc : int -> int_vector = "caml_int_vector_alloc"
-external int_vector_length : int_vector -> int = "caml_int_vector_length"
-external int_vector_set : int_vector -> int -> int -> unit = "caml_int_vector_set"
-
-external tree_select_child : tree -> [`Tree ] node -> int_vector -> [`Tree] node = "caml_xml_tree_select_child"
-external tree_select_foll_sibling : tree -> [`Tree ] node -> int_vector -> [`Tree] node = "caml_xml_tree_select_foll_sibling"
-external tree_select_desc : tree -> [`Tree ] node -> int_vector -> [`Tree] node = "caml_xml_tree_select_desc"
-external tree_select_foll_below : tree -> [`Tree ] node -> int_vector -> [`Tree] node -> [`Tree] node = "caml_xml_tree_select_foll_below"
+external tree_select_child : tree -> [`Tree ] node -> unordered_set -> [`Tree] node = "caml_xml_tree_select_child" "noalloc"
+external tree_select_foll_sibling : tree -> [`Tree ] node -> unordered_set -> [`Tree] node = "caml_xml_tree_select_foll_sibling" "noalloc"
+external tree_select_desc : tree -> [`Tree ] node -> unordered_set -> [`Tree] node = "caml_xml_tree_select_desc" "noalloc"
+external tree_select_foll_below : tree -> [`Tree ] node -> unordered_set -> [`Tree] node -> [`Tree] node = "caml_xml_tree_select_foll_below" "noalloc"
module HPtset = Hashtbl.Make(Ptset.Int)
HPtset.find vector_htbl s
with
Not_found ->
- let v = int_vector_alloc (Ptset.Int.cardinal s) in
- let _ = Ptset.Int.fold (fun e i -> int_vector_set v i e;i+1) s 0 in
+ let v = unordered_set_alloc (Ptset.Int.cardinal s) in
+ let _ = Ptset.Int.iter (fun e -> unordered_set_insert v e) s in
HPtset.add vector_htbl s v; v
type t = {
doc : tree;
- ttable : (Tag.t,(Ptset.Int.t*Ptset.Int.t)) Hashtbl.t;
+ ttable : (Tag.t,(Ptset.Int.t*Ptset.Int.t*Ptset.Int.t*Ptset.Int.t)) Hashtbl.t;
}
-
+let subtree_size t i = tree_subtree_size t.doc i
let text_size t = text_size t.doc
module MemUnion = Hashtbl.Make (struct
if x < y then HASHINT2(x,y) else HASHINT2(y,x)
end)
+module MemAdd = Hashtbl.Make (
+ struct
+ type t = Tag.t*Ptset.Int.t
+ let equal (x,y) (z,t) = (x == z)&&(Ptset.Int.equal y t)
+ let hash (x,y) = HASHINT2(x,Ptset.Int.hash y)
+ end)
+
let collect_tags tree =
let h_union = MemUnion.create BIG_H_SIZE in
let pt_cup s1 s2 =
in
MemUnion.add h_union (s1,s2) s;s
in
- let h_add = Hashtbl.create BIG_H_SIZE in
- let pt_add t s =
- let k = HASHINT2(Tag.hash t,Ptset.Int.hash s) in
- try
- Hashtbl.find h_add k
- with
+ let h_add = MemAdd.create BIG_H_SIZE in
+ let pt_add t s =
+ try MemAdd.find h_add (t,s)
+ with
| Not_found -> let r = Ptset.Int.add t s in
- Hashtbl.add h_add k r;r
+ MemAdd.add h_add (t,s) r;r
in
let h = Hashtbl.create BIG_H_SIZE in
- let update t sb sa =
- let sbelow,safter =
+ let update t sc sb ss sa =
+ let schild,sbelow,ssibling,safter =
try
Hashtbl.find h t
with
| Not_found ->
- (Ptset.Int.empty,Ptset.Int.empty)
+ (Ptset.Int.empty,Ptset.Int.empty,Ptset.Int.empty,Ptset.Int.empty)
in
- Hashtbl.replace h t (pt_cup sbelow sb, pt_cup safter sa)
+ Hashtbl.replace h t
+ (pt_cup sc schild,pt_cup sbelow sb, pt_cup ssibling ss, pt_cup safter sa)
in
- let rec loop id acc =
- if equal_node id nil
- then (Ptset.Int.empty,acc)
+ let rec loop_right id acc_sibling acc_after=
+ if id == nil
+ then (acc_sibling,acc_after)
+ else
+ let sibling2,after2 = loop_right (tree_next_sibling tree id) acc_sibling acc_after in
+ let child1,below1 = loop_left (tree_first_child tree id) after2 in
+ let tag = tree_tag_id tree id in
+ update tag child1 below1 sibling2 after2;
+ (pt_add tag sibling2, (pt_add tag (pt_cup after2 below1)))
+ and loop_left id acc_after =
+ if id == nil
+ then (Ptset.Int.empty,Ptset.Int.empty)
else
- let below2,after2 = loop (tree_next_sibling tree id) acc in
- let below1,after1 = loop (tree_first_child tree id) after2 in
+ let sibling2,after2 = loop_right (tree_next_sibling tree id) Ptset.Int.empty acc_after in
+ let child1,below1 = loop_left (tree_first_child tree id) after2 in
let tag = tree_tag_id tree id in
- update tag below1 after2;
- pt_add tag (pt_cup below1 below2), (pt_add tag after1)
+ update tag child1 below1 sibling2 after2;
+ (pt_add tag sibling2,(pt_add tag (pt_cup after2 below1)))
in
- let _ = loop (tree_root tree) Ptset.Int.empty in h
-
-
-
-
+ let _ = loop_left (tree_root tree) Ptset.Int.empty in h
+
+
+
let contains_array = ref [| |]
let contains_index = Hashtbl.create 4096
let s = text_get_cached_text t.doc n
in
if matching s
- then loop (n+1) (n::acc) (l+1)
- else loop (n+1) acc l
+ then loop (nodei ((inode n)+1)) (n::acc) (l+1)
+ else loop (nodei ((inode n)+1)) acc l
in
let acc,l = loop i [] 0 in
- let a = Array.create l nil in
+ let a = Array.create l nulldoc in
let _ = List.fold_left (fun cpt e -> a.(cpt) <- e; (cpt-1)) (l-1) acc
in
contains_array := a
+
+let last_idx = ref 0
+
+let array_find a i j =
+ let l = Array.length a in
+ let rec loop idx x y =
+ if x > y || idx >= l then nulldoc
+ else
+ if a.(idx) >= x then if a.(idx) > y then nulldoc else (last_idx := idx;a.(idx))
+ else loop (idx+1) x y
+ in
+ if a.(0) > j || a.(l-1) < i then nulldoc
+ else loop !last_idx i j
+let text_below tree t =
+ let l = Array.length !contains_array in
+ let i,j = tree_doc_ids tree.doc t in
+ let id = if l == 0 then i else (array_find !contains_array i j) in
+ tree_parent_doc tree.doc id
+
+let text_next tree t root =
+ let l = Array.length !contains_array in
+ let inf = nodei((inode(snd(tree_doc_ids tree.doc t)))+1) in
+ let _,j = tree_doc_ids tree.doc root in
+ let id = if l == 0 then if inf > j then nulldoc else inf
+ else array_find !contains_array inf j
+ in
+ tree_parent_doc tree.doc id
+
module DocIdSet = struct
external pool : tree -> Tag.pool = "%identity"
-let save t str = (save_tree t.doc str)
+let magic_string = "SXSI_INDEX"
+let version_string = "1"
+
+let pos fd =
+ Unix.lseek fd 0 Unix.SEEK_CUR
+
+let pr_pos fd = Printf.eprintf "At position %i\n%!" (pos fd)
+
+let write fd s =
+ let sl = String.length s in
+ let ssl = Printf.sprintf "%020i" sl in
+ ignore (Unix.write fd ssl 0 20);
+ ignore (Unix.write fd s 0 (String.length s))
+
+let rec really_read fd buffer start length =
+ if length <= 0 then () else
+ match Unix.read fd buffer start length with
+ 0 -> raise End_of_file
+ | r -> really_read fd buffer (start + r) (length - r);;
+
+let read fd =
+ let buffer = String.create 20 in
+ let _ = really_read fd buffer 0 20 in
+ let size = int_of_string buffer in
+ let buffer = String.create size in
+ let _ = really_read fd buffer 0 size in
+ buffer
+
+
+let save t str =
+ let fd = Unix.openfile str [ Unix.O_WRONLY;Unix.O_TRUNC;Unix.O_CREAT] 0o644 in
+ let out_c = Unix.out_channel_of_descr fd in
+ let _ = set_binary_mode_out out_c true in
+ output_string out_c magic_string;
+ output_char out_c '\n';
+ output_string out_c version_string;
+ output_char out_c '\n';
+ Marshal.to_channel out_c t.ttable [ ];
+ (* we need to move the fd to the correct position *)
+ flush out_c;
+ ignore (Unix.lseek fd (pos_out out_c) Unix.SEEK_SET);
+ tree_save t.doc fd;
+ close_out out_c
;;
let load ?(sample=64) str =
- node_of_t (load_tree str sample)
-
+ let fd = Unix.openfile str [ Unix.O_RDONLY ] 0o644 in
+ let in_c = Unix.in_channel_of_descr fd in
+ let _ = set_binary_mode_in in_c true in
+ (let ms = input_line in_c in if ms <> magic_string then failwith "Invalid index file");
+ (let vs = input_line in_c in if vs <> version_string then failwith "Invalid version file");
+ let table : (Tag.t,(Ptset.Int.t*Ptset.Int.t*Ptset.Int.t*Ptset.Int.t)) Hashtbl.t =
+ Marshal.from_channel in_c
+ in
+ let ntable = Hashtbl.create (Hashtbl.length table) in
+ Hashtbl.iter (fun k (s1,s2,s3,s4) ->
+ let ss1 = Ptset.Int.fold (Ptset.Int.add) s1 Ptset.Int.empty
+ and ss2 = Ptset.Int.fold (Ptset.Int.add) s2 Ptset.Int.empty
+ and ss3 = Ptset.Int.fold (Ptset.Int.add) s3 Ptset.Int.empty
+ and ss4 = Ptset.Int.fold (Ptset.Int.add) s4 Ptset.Int.empty
+ in Hashtbl.add ntable k (ss1,ss2,ss3,ss4)
+ ) table;
+ Hashtbl.clear table;
+ (* The in_channel read a chunk of fd, so we might be after
+ the start of the XMLTree save file. Reset to the correct
+ position *)
+ ignore(Unix.lseek fd (pos_in in_c) Unix.SEEK_SET);
+ let tree = { doc = tree_load fd;
+ ttable = ntable;}
+ in close_in in_c;
+ tree
+
let tag_pool t = pool t.doc
-let compare a b = a - b
+let compare = compare_node
let equal a b = a == b
-1 -> "Nil"
| i -> Printf.sprintf "Node (%i)" i
-let dump_node t = nts t
+let dump_node t = nts (inode t)
-
let is_left t n = tree_is_first_child t.doc n
-let is_below_right t n1 n2 = tree_is_ancestor t.doc (tree_parent t.doc n1) n2
-
+let is_below_right t n1 n2 =
+ tree_is_ancestor t.doc (tree_parent t.doc n1) n2
+ && not (tree_is_ancestor t.doc n1 n2)
+
let parent t n = tree_parent t.doc n
let first_child t = (); fun n -> tree_first_child t.doc n
+let first_element t = (); fun n -> tree_first_element t.doc n
(* these function will be called in two times: first partial application
on the tag, then application of the tag and the tree, then application of
fun n -> tree_select_child t.doc n v
let next_sibling t = (); fun n -> tree_next_sibling t.doc n
+let next_element t = (); fun n -> tree_next_element t.doc n
+
let tagged_sibling t tag = (); fun n -> tree_tagged_sibling t.doc n tag
let select_sibling t = fun ts ->
fun n -> tree_select_foll_sibling t.doc n v
let next_sibling_ctx t = (); fun n _ -> tree_next_sibling t.doc n
+let next_element_ctx t = (); fun n _ -> tree_next_element t.doc n
let tagged_sibling_ctx t tag = (); fun n _ -> tree_tagged_sibling t.doc n tag
let select_sibling_ctx t = fun ts ->
let v = (ptset_to_vector ts) in ();
fun n -> tree_select_desc t.doc n v
-let tagged_foll_ctx t tag = (); fun n ctx -> tree_tagged_foll_below t.doc n tag ctx
+let tagged_foll_ctx t tag = (); fun n ctx -> tree_tagged_foll_below t.doc n tag ctx
let select_foll_ctx t = fun ts ->
let v = (ptset_to_vector ts) in ();
then
let tagid = tree_tag_id tree.doc t in
if tagid==Tag.pcdata
- then output_string outc (text_get_cached_text tree.doc t);
- if print_right
- then loop (next_sibling tree t)
-
+ then
+ begin
+ let tid = tree_my_text tree.doc t in
+ let _ = Printf.eprintf "my_text %i returned %i\n%!" (inode t) (inode tid)
+ in
+ output_string outc (text_get_cached_text tree.doc tid);
+ if print_right
+ then loop (next_sibling tree t);
+ end
else
let tagstr = Tag.to_string tagid in
let l = first_child tree t
output_char outc '>';
end;
if print_right then loop r
- and loop_attributes a =
+ and loop_attributes a =
+ if a != nil
+ then
let s = (Tag.to_string (tag tree a)) in
let attname = String.sub s 3 ((String.length s) -3) in
+ let fsa = first_child tree a in
+ let tid = tree_my_text tree.doc fsa in
+ let _ = Printf.eprintf "my_text %i returned %i\n%!" (inode fsa) (inode tid)
+ in
output_char outc ' ';
output_string outc attname;
output_string outc "=\"";
- output_string outc (text_get_cached_text tree.doc
- (tree_my_text tree.doc (first_child tree a)));
+ output_string outc (text_get_cached_text tree.doc tid);
output_char outc '"';
loop_attributes (next_sibling tree a)
in
print_xml_fast outc tree (first_child tree t)
else print_xml_fast outc tree t
+let tags_children t tag =
+ let a,_,_,_ = Hashtbl.find t.ttable tag in a
let tags_below t tag =
- fst(Hashtbl.find t.ttable tag)
-
+ let _,a,_,_ = Hashtbl.find t.ttable tag in a
+let tags_siblings t tag =
+ let _,_,a,_ = Hashtbl.find t.ttable tag in a
let tags_after t tag =
- snd(Hashtbl.find t.ttable tag)
+ let _,_,_,a = Hashtbl.find t.ttable tag in a
+
let tags t tag = Hashtbl.find t.ttable tag
-let binary_parent t n =
+let rec binary_parent t n =
+ let r =
if tree_is_first_child t.doc n
then tree_parent t.doc n
else tree_prev_sibling t.doc n
+ in if tree_tag_id t.doc r = Tag.pcdata then
+ binary_parent t r
+ else r
let doc_ids t n = tree_doc_ids t.doc n
let get_text t n =
let tid = tree_my_text t.doc n in
- if tid == nil then "" else
+ if tid == nulldoc then "" else
text_get_cached_text t.doc tid
projects / SXSI / xpathcomp.git / blobdiff